A deep learning-based vehicle energy consumption performance evaluation generation method and device

Abstract

The application discloses a kind of vehicle energy consumption performance evaluation generation method and device based on deep learning, it is related to vehicle market analysis field, the method collects vehicle operation data, extracts energy consumption related quantitative data, statistics working condition proportion characteristic, obtains model input characteristic data after standardization processing;With pre-processing data and target comment sample constructs training data set, uses supervised learning to combine loss function optimization model parameter, completes model training, inputs energy consumption data into the model trained, and outputs vehicle energy consumption grade and the sorting result of energy consumption influence factor with weight;Rely on the preset multi-scene energy consumption analysis scene library and comment template, match energy consumption grade and working condition weight, embed quantitative data and sorting result, automatically generate vehicle energy consumption performance evaluation.The present application can realize objective, accurate, efficient vehicle energy consumption intelligent evaluation.

Description

Technical Field

[0001] This invention relates to the field of vehicle market analysis, and in particular to a method for generating vehicle energy consumption performance evaluation based on deep learning, a device for generating vehicle energy consumption performance evaluation based on deep learning, an electronic device, and a storage medium. Background Technology

[0002] Currently, with the rapid development of vehicle-to-everything (V2X) big data technology, the potential value within this data is gradually being discovered and applied. Accurate and timely assessment of commercial vehicle energy consumption using V2X data is crucial for commercial vehicle manufacturers to optimize product strategies, dealers to formulate sales plans, and investors to make informed decisions. Traditional assessment methods often rely on manual analysis of large amounts of statistical data to write evaluations. This process is not only time-consuming and labor-intensive but also susceptible to subjective factors, resulting in inconsistent and uneven quality of evaluations.

[0003] With the rise of deep learning technology, leveraging its powerful data processing and text generation capabilities to automatically generate comments has become an effective way to improve the efficiency and accuracy of evaluations. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a deep learning-based method for generating vehicle energy consumption performance evaluations, a deep learning-based device for generating vehicle energy consumption performance evaluations, an electronic device, and a storage medium. The aim is to solve the technical problem that the existing technology relies heavily on manual analysis of a large amount of statistical data to write evaluations. This process is not only time-consuming and laborious, but also easily affected by subjective factors, resulting in inconsistent quality and poor consistency of the evaluations.

[0005] This invention provides the following solution:

[0006] According to one aspect of the present invention, a method for generating vehicle energy consumption performance evaluation based on deep learning is provided, comprising the following steps:

[0007] Collect vehicle operation data and extract energy consumption-related quantitative data from the vehicle operation data;

[0008] The operating condition ratio features of the vehicle operation data are obtained, and the operating condition ratio features are standardized sequentially to obtain the model input feature data.

[0009] Construct a pre-defined comment generation model;

[0010] Obtain target comment samples; associate preprocessed data with the target comment samples to construct a training dataset;

[0011] The training dataset is input into the preset comment generation model. Supervised learning is used in combination with the loss function to optimize the model parameters. The generalization ability of the model is evaluated through validation methods to complete the model training.

[0012] Input the energy consumption-related quantitative data into the pre-trained evaluation generation model, and output the vehicle energy consumption level results and the weighted ranking results of energy consumption influencing factors.

[0013] A library of pre-defined energy consumption analysis scenarios categorized by vehicle driving conditions, with each scenario associated with a different energy consumption level and a comment template.

[0014] Based on the energy consumption level and operating condition weights output by the model, a corresponding evaluation template is matched, and quantitative data and the ranking results of energy consumption influencing factors are embedded to generate an energy consumption performance evaluation result.

[0015] Furthermore, energy consumption-related quantitative data includes: fuel consumption data, vehicle speed data, road type data, and terrain data.

[0016] The operating condition proportion is the proportion of driving mileage corresponding to each type of road and terrain to the total driving mileage.

[0017] Furthermore, the pre-defined comment generation model is as follows:

[0018] A model combining long short-term memory networks with attention mechanisms;

[0019] The model includes:

[0020] Input layer: Converts the input feature data of the model into multi-dimensional feature vectors and inputs them into the model;

[0021] Feature extraction layer: processes the input feature data, captures the correlation between data, extracts temporal features from the data, and identifies and analyzes potential patterns;

[0022] Key Feature Focusing Layer: The output features of the feature extraction layer are weighted and fused, and the weights are dynamically assigned according to the importance of the features to the generation of energy consumption comments, so as to realize the automatic focusing of key factors affecting energy consumption.

[0023] Feature integration layer: Through linear transformation combined with nonlinear activation function, the weighted and fused features are nonlinearly combined and dimensionally mapped to extract high-dimensional abstract features;

[0024] Probabilistic output layer: A normalized probabilistic activation function is used to convert the output of the feature integration layer into a probability distribution, providing a probabilistic output basis for model training and inference.

[0025] Furthermore, including:

[0026] The loss value during model training is calculated using a cross-entropy loss function.

[0027] The model parameters are adjusted using gradient descent-type optimization algorithms. When the loss value tends to stabilize and the model's generalization ability meets the preset requirements, the model training is completed.

[0028] Furthermore, including:

[0029] The vehicle energy consumption rating results are multi-level quantitative ratings, and the ranking results of energy consumption influencing factors include the weight ratio and priority relationship of each influencing factor.

[0030] Furthermore, including:

[0031] The driving condition combinations in the scenario library are combinations of road types and terrain distributions. The comment templates reserve placeholders for quantitative information, which includes numerical data related to energy consumption, operating condition ratios, and driving behavior.

[0032] Furthermore, the generated energy consumption performance evaluation of the commercial vehicle includes:

[0033] The content includes energy consumption levels, driving conditions, quantitative statistics, factors affecting energy consumption, and optimization suggestions.

[0034] According to a second aspect of the present invention, a deep learning-based vehicle energy consumption performance evaluation generation device is provided, comprising:

[0035] The system includes modules for data acquisition and extraction, data preprocessing, model building, dataset building, model training, energy consumption analysis, scenario library, and evaluation result generation.

[0036] The data acquisition and extraction module is used to collect vehicle operation data and extract energy consumption-related quantitative data from the vehicle operation data.

[0037] The data preprocessing module is used to statistically analyze the operating condition proportion characteristics of each data item, and then performs noise and outlier cleaning and dimension normalization on the statistically analyzed data to obtain the feature data for model input.

[0038] The model building module is used to build a preset comment generation model;

[0039] The dataset construction module is used to obtain target comment samples; it associates preprocessed data with target comment samples to construct a training dataset.

[0040] The model training module is used to input the training dataset into the preset comment generation model, optimize the model parameters by using supervised learning combined with the loss function, and evaluate the model's generalization ability through validation methods to complete the model training.

[0041] The energy consumption analysis module is used to input energy consumption-related quantitative data into a pre-trained pre-defined rating generation model and output vehicle energy consumption level results and weighted ranking results of energy consumption influencing factors.

[0042] The scenario library module is used to preset energy consumption analysis scenario libraries divided according to vehicle driving conditions, and each scenario is associated with different energy consumption level evaluation templates.

[0043] The evaluation result generation module is used to match the corresponding comment template based on the energy consumption level and operating condition weight output by the model, embed quantitative data and the ranking results of energy consumption influencing factors, and generate energy consumption performance evaluation results.

[0044] According to three aspects of the present invention, an electronic device is provided, comprising: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0045] The memory stores a computer program, which, when executed by a processor, causes the processor to perform steps of a deep learning-based method for generating vehicle energy consumption performance evaluations.

[0046] According to four aspects of the present invention, a computer-readable storage medium is provided that stores a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of a deep learning-based method for generating vehicle energy consumption performance evaluation.

[0047] Compared with the prior art, the present invention has the following advantages:

[0048] This application automatically, quickly, and accurately generates comprehensive, objective, and targeted evaluations of commercial vehicle market performance based on statistical data such as fuel consumption, mileage percentage of vehicle speed, mileage percentage of road type distribution (highway, city, suburbs), and mileage percentage of terrain distribution (plains, mountains, hills), providing strong support for relevant decision-making.

[0049] This application delves into the value of connected vehicle big data, meticulously categorizes various factors affecting the energy consumption performance of commercial vehicles, extracts and standardizes key statistical data required for each category, and finally outputs high-quality and consistent comments based on a deep learning model, helping development and sales personnel quickly locate problems and promote products in the market. Attached Figure Description

[0050] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0051] Figure 1 This is a flowchart of a method for generating vehicle energy consumption performance evaluation based on deep learning, provided by one or more embodiments of the present invention.

[0052] Figure 2 This is a structural diagram of a vehicle energy consumption performance evaluation generation device based on deep learning, provided by one or more embodiments of the present invention.

[0053] Figure 3 This is a block diagram of an electronic device structure for a method for generating vehicle energy consumption performance evaluation based on deep learning, provided by one or more embodiments of the present invention. Detailed Implementation

[0054] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the application. The singular forms “a,” “said,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0056] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0057] It should be understood that although the terms first, second, third, etc., may be used in the embodiments of this application, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, first may also be referred to as second without departing from the scope of the embodiments of this application, and similarly, second may also be referred to as first.

[0058] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”

[0059] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that an article or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the article or device that includes said element.

[0060] It should be noted that any symbols and / or numbers present in the specification that are not marked in the accompanying drawings are not reference numerals.

[0061] Figure 1 This is a flowchart of a method for generating vehicle energy consumption performance evaluation based on deep learning, provided by one or more embodiments of the present invention.

[0062] like Figure 1 As shown, it includes the following steps:

[0063] Step S1: Collect vehicle operation data and extract energy consumption-related quantitative data from the vehicle operation data;

[0064] Specifically, energy consumption-related quantitative data include: fuel consumption data, vehicle speed data, road type data, and terrain data.

[0065] The operating condition proportion is the proportion of driving mileage corresponding to each type of road and terrain to the total driving mileage.

[0066] Step S2: Statistically analyze the operating condition proportion characteristics of each data item, and then perform noise and outlier cleaning and dimension normalization on the statistically analyzed data to obtain the feature data for model input.

[0067] Step S3: Construct a preset comment generation model;

[0068] Specifically, the preset comment generation model is as follows:

[0069] A model combining long short-term memory networks with attention mechanisms;

[0070] The model includes:

[0071] Input layer: Converts the input feature data of the model into multi-dimensional feature vectors and inputs them into the model;

[0072] Feature extraction layer: Processes sequence feature data through a gating mechanism, captures the correlation between data, and extracts temporal features and potential patterns from the data;

[0073] Key Feature Focusing Layer: The output features of the feature extraction layer are weighted and fused, and the weights are dynamically assigned according to the importance of the features to the generation of energy consumption comments, so as to realize the automatic focusing of key factors affecting energy consumption.

[0074] Feature integration layer: Through linear transformation combined with nonlinear activation function, the weighted and fused features are nonlinearly combined and dimensionally mapped to extract high-dimensional abstract features;

[0075] Probabilistic output layer: A normalized probabilistic activation function is used to convert the output of the feature integration layer into a probability distribution, providing a probabilistic output basis for model training and inference.

[0076] Step S4: Obtain target comment samples; associate the preprocessed data with the target comment samples to construct a training dataset;

[0077] Step S5: Input the training dataset into the preset comment generation model, optimize the model parameters by using supervised learning combined with the loss function, and evaluate the model's generalization ability through validation methods to complete the model training.

[0078] Specifically, the cross-entropy loss function is used to calculate the loss value during model training;

[0079] The model parameters are adjusted using gradient descent-type optimization algorithms. When the loss value tends to stabilize and the model's generalization ability meets the preset requirements, the model training is completed.

[0080] Step S6: Input the energy consumption-related quantitative data into the trained preset comment generation model, and output the vehicle energy consumption level results and the weighted ranking results of energy consumption influencing factors.

[0081] Step S7: Preset an energy consumption analysis scenario library divided according to vehicle driving conditions, and associate each scenario with a comment template of different energy consumption levels.

[0082] Step S8: Match the corresponding comment template according to the energy consumption level and operating condition weight output by the model, embed the quantitative data and the ranking results of energy consumption influencing factors, and generate the energy consumption performance evaluation results.

[0083] Specifically, the generated energy consumption performance evaluation for the commercial vehicle includes:

[0084] The content includes energy consumption levels, driving conditions, quantitative statistics, factors affecting energy consumption, and optimization suggestions.

[0085] Energy consumption levels are divided into multiple quantitative levels, and the ranking results of energy consumption influencing factors include the weight ratio and priority relationship of each influencing factor.

[0086] The driving condition combinations in the scenario library are combinations of road types and terrain distributions. The comment templates reserve placeholders for quantitative information, which includes numerical data related to energy consumption, operating condition ratios, and driving behavior.

[0087] Specifically, by collecting multi-source vehicle operation data such as fuel consumption, vehicle speed, road type, and terrain, the characteristics of operating conditions are extracted to fully characterize various influencing factors of vehicle energy consumption, making the evaluation results more consistent with actual driving scenarios and improving the accuracy and comprehensiveness of the evaluation.

[0088] By sequentially cleaning noise and outliers and normalizing dimensions for the operating condition proportion characteristics, data interference and dimensional differences are effectively eliminated, improving the quality of model input data, thereby accelerating the convergence speed of subsequent deep learning models and ensuring the stability and reliability of energy consumption evaluation results.

[0089] By constructing a comment generation model using a long short-term memory network combined with an attention mechanism, the model can capture the temporal dependence and potential patterns of operational data through the long short-term memory network, and dynamically allocate feature weights through the attention mechanism. This enables the automatic focusing and weighted ranking of key energy consumption influencing factors, solving the technical problem of traditional methods being unable to quantify key influencing factors.

[0090] A training dataset is constructed by associating target evaluation samples with preprocessed data. A supervised learning approach is adopted, and the model parameters are iteratively optimized by combining cross-entropy loss function and gradient descent optimization algorithm to ensure that the model can play a stable role under different vehicle models and driving conditions, and has a wide range of applications.

[0091] The evaluation results are interpretable and practical. They not only output the vehicle energy consumption level, but also provide a weighted ranking of energy consumption influencing factors, providing clear decision-making guidance for drivers and fleet managers. This facilitates targeted improvements in driving behavior and strategies, thereby optimizing vehicle energy consumption and demonstrating high practical application value.

[0092] Figure 2 This is a structural diagram of a vehicle energy consumption performance evaluation generation device based on deep learning, provided by one or more embodiments of the present invention.

[0093] like Figure 2 As shown, it includes:

[0094] The system includes modules for data acquisition and extraction, data preprocessing, model building, dataset building, model training, energy consumption analysis, scenario library, and evaluation result generation.

[0095] The data acquisition and extraction module is used to collect vehicle operation data and extract energy consumption-related quantitative data from the vehicle operation data.

[0096] The data preprocessing module is used to statistically analyze the operating condition proportion characteristics of each data item, and then performs noise and outlier cleaning and dimension normalization on the statistically analyzed data to obtain the feature data for model input.

[0097] The model building module is used to build a preset comment generation model;

[0098] The dataset construction module is used to obtain target comment samples; it associates preprocessed data with target comment samples to construct a training dataset.

[0099] The model training module is used to input the training dataset into the preset comment generation model, optimize the model parameters by using supervised learning combined with the loss function, and evaluate the model's generalization ability through validation methods to complete the model training.

[0100] The energy consumption analysis module is used to input energy consumption-related quantitative data into a pre-trained pre-defined rating generation model and output vehicle energy consumption level results and weighted ranking results of energy consumption influencing factors.

[0101] The scenario library module is used to preset energy consumption analysis scenario libraries divided according to vehicle driving conditions, and each scenario is associated with different energy consumption level evaluation templates.

[0102] The evaluation result generation module is used to match the corresponding comment template based on the energy consumption level and operating condition weight output by the model, embed quantitative data and the ranking results of energy consumption influencing factors, and generate energy consumption performance evaluation results.

[0103] It is worth noting that although only some basic functional modules are disclosed in this embodiment, it does not mean that the composition of this system is limited to the above-mentioned basic functional modules. On the contrary, what this embodiment intends to express is that, based on the above-mentioned basic functional modules, those skilled in the art can arbitrarily add one or more functional modules in combination with existing technology to form an infinite number of embodiments or technical solutions. That is to say, this system is open rather than closed. The fact that this embodiment only discloses a few basic functional modules does not mean that the scope of protection of the claims of this invention is limited to the disclosed basic functional modules. At the same time, for the convenience of description, the above device is described separately according to its functions as various units and modules. Of course, in implementing this invention, the functions of each unit and module can be implemented in one or more software and / or hardware.

[0104] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0105] In one embodiment, the following steps are included:

[0106] In this embodiment, the on-board telematics terminal (tbox system) built into the commercial vehicle collects multi-dimensional operating data in real time during the vehicle's operation. The collected operating data includes at least fuel consumption data, vehicle speed data, road type data, and terrain data. The road type data includes three categories: highway driving data, urban road driving data, and suburban road driving data. The terrain data includes three categories: plain terrain driving data, mountainous terrain driving data, and hilly terrain driving data.

[0107] The collected vehicle operation data is accumulated periodically, and the proportion of mileage corresponding to each type of data is statistically analyzed based on the total mileage. The proportions of mileage on highways, urban roads, and suburban roads to the total mileage, as well as the proportions of mileage on plains, mountains, and hills to the total mileage, are obtained, thus completing the statistical processing of vehicle operation data.

[0108] Data cleaning:

[0109] After statistical processing, the vehicle operation data is cleaned for noise and outliers to remove noisy data and outliers. Specifically, for fuel consumption data, values ​​that exceed the normal range of fuel consumption under similar driving conditions for the same type of vehicle are identified as outliers and removed to ensure the validity and accuracy of the vehicle operation data input into the subsequent model.

[0110] Data standardization:

[0111] After the vehicle operation data has been cleaned, a dimensional normalization process is performed to eliminate the dimensional differences between different types of data and make the data in each dimension comparable. Specifically, for vehicle operation data with different dimensions such as vehicle speed data and fuel consumption data, a normalization process is used to map each data value to the [0,1] interval to obtain standardized feature data, so as to adapt to the input processing requirements of the subsequent deep learning model.

[0112] Building deep learning models:

[0113] A comment generation model is constructed by using an LSTM (Long Short-Term Memory) network combined with an attention mechanism. The model mainly consists of an input layer, an LSTM layer, an attention layer, a fully connected layer, and an output layer.

[0114] The input layer receives preprocessed statistical data on commercial vehicles. This data is input into the model in vector form, where each dimension corresponds to a statistical feature, such as standardized fuel consumption, percentage of mileage on highways, and percentage of mileage in plains areas. The role of the input layer is to introduce the raw data into the model, providing a foundation for subsequent processing.

[0115] LSTM Layer: LSTM is a special type of recurrent neural network (RNN) that effectively handles long-term dependencies in sequential data. In the task of generating performance reviews for the commercial vehicle market, different statistical data may have certain temporal or logical correlations; for example, changes in vehicle speed may be related to road type and terrain. LSTM layers control the flow of information through their unique gating mechanisms (input gate, forget gate, and output gate), enabling them to remember important historical information and forget irrelevant information. In this model, the LSTM layer processes the data from the input layer step by step, extracting temporal features and potential patterns from the data. For example, it can learn the changing patterns of vehicle speed and fuel consumption under different terrains and road types.

[0116] Attention Layer: The attention mechanism is a technique that allows the model to automatically focus on important parts of the input data. After the LSTM processes the data, the attention layer performs a weighted sum of the features at each time step of the LSTM output. The weight reflects the importance of the feature at that time step to generating the evaluation. For example, if a set of statistics shows a high proportion of mileage driven in mountainous terrain, then features related to mountain driving (such as fuel consumption and speed in mountainous areas) may be given higher weights in the attention layer, because these features are more crucial for generating evaluations of the commercial vehicle's performance in the mountainous market. Through the attention mechanism, the model can focus more on the information most relevant to the current task, improving the accuracy and relevance of the generated evaluations.

[0117] Fully connected layers: Fully connected layers further transform and integrate the weighted feature vectors output by the attention layer. They non-linearly combine features through a series of linear transformations and non-linear activation functions (such as ReLU), mapping high-dimensional feature vectors to a lower-dimensional space while extracting higher-level abstract features. These abstract features are more conducive to the model generating accurate and concise comments.

[0118] Output Layer: The output layer uses the softmax activation function to transform the vector output by the fully connected layer into a probability distribution. Each probability value corresponds to a possible comment word. The model selects the word with the highest probability as the output at the current time step, gradually generating a complete comment. For example, when generating the first word of a comment, the output layer calculates the probability of each possible starting word based on the feature vector from the fully connected layer, selects the word with the highest probability as the beginning of the comment, and so on, until a complete comment sentence is generated.

[0119] Data Preparation: Collect a large number of comments related to the performance of the commercial vehicle market, and associate the above statistical data with the corresponding comments to construct a training dataset. For example, a training sample is composed of a set of data such as fuel consumption, speed-to-mileage ratio, road type, and terrain distribution-to-mileage ratio, along with a comment describing the market performance of the commercial vehicle.

[0120] Training Process: The training dataset is input into the constructed LSTM + attention mechanism model, and supervised learning is used for training. During training, the gradient of the loss function (such as the cross-entropy loss function) with respect to the model parameters is calculated using the backpropagation algorithm. Then, the model parameters are gradually adjusted using optimization algorithms (such as stochastic gradient descent or its improved versions) to continuously reduce the loss function value. Through continuous iterative training, the model learns the mapping relationship between statistical data and comments, thus gaining the ability to generate corresponding comments based on the input statistical data. Simultaneously, methods such as cross-validation are used to evaluate the model's performance, ensuring that the model has good generalization ability and can generate accurate and reasonable comments on unseen data.

[0121] A new set of preprocessed commercial vehicle statistics (fuel consumption, mileage percentage by speed, mileage percentage by road type, and mileage percentage by terrain) is input into the trained deep learning model. Based on the input data and the knowledge learned during training, the model automatically generates the corresponding commercial vehicle market energy consumption level (high, medium, low) and the ranking of influencing factors (e.g., road type, terrain).

[0122] Scene library construction:

[0123] A pre-set comment template library is provided, and scenarios are divided according to road type and terrain distribution (such as "highway-plain" and "city-mountain"). Each scenario is associated with multiple comment templates.

[0124] (For example: Highway-Plains-High Fuel Consumption: "The vehicle's fuel consumption is high in highway and plains scenarios (XX L / 100km). It is recommended to check the engine load or tire pressure." City-Mountain-Medium Fuel Consumption: "The vehicle's fuel consumption is moderate in city and mountain scenarios (XX L / 100km), but there are many times of rapid acceleration (XX times / 100km). Driving habits need to be optimized.")

[0125] Comment generation rules:

[0126] Based on the fuel consumption classification and scene weights output by the model, match the corresponding templates in the scene library;

[0127] Insert the priority and percentage of influencing factors in the model output (e.g., "mountainous terrain accounts for 45%, which is the primary influencing factor");

[0128] Generate the final evaluation (e.g.: "The vehicle has moderate fuel consumption in urban and mountainous scenarios (28L / 100km), but the mountainous terrain accounts for 45% of the total, resulting in a high load, and the number of rapid accelerations is 30 times / 100km (accounting for 25%), which needs to be optimized.").

[0129] Specifically, by using an attention mechanism to calculate the weights of different features (such as the proportion of mountainous terrain and the number of rapid accelerations) in parallel, the system achieves dynamic focusing on key influencing factors, avoiding the subjectivity of manually setting weights. It outputs a priority list of influencing factors (e.g., "Mountainous terrain: 45% > Rapid acceleration: 30%)", enhancing the interpretability of the evaluation.

[0130] By decoupling the scene library from the model output, the comment generation is divided into two independent modules: model inference (outputting fuel consumption classification and influencing factor weights) and scene library matching (inserting specific scene templates). This supports flexible expansion and updates of the scene library. It also reduces model iteration costs; adding new scenes (such as "suburban-hilly") only requires updating the template library, without retraining the model.

[0131] Figure 3 This is a block diagram of an electronic device structure for a lane line recognition method based on multi-feature fusion provided by one or more embodiments of the present invention.

[0132] like Figure 3 As shown, this application provides an electronic device, including: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0133] The memory stores a computer program that, when executed by a processor, causes the processor to perform steps of a lane line recognition method based on multi-feature fusion.

[0134] This application also provides a computer-readable storage medium storing a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform steps of a lane line recognition method based on multi-feature fusion.

[0135] For the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0136] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating vehicle energy consumption performance evaluation based on deep learning, characterized in that, Collect vehicle operation data and extract energy consumption-related quantitative data from the vehicle operation data; The operating condition ratio features of the vehicle operation data are obtained, and the operating condition ratio features are standardized sequentially to obtain the model input feature data. Construct a pre-defined comment generation model; Obtain target comment samples; associate the input feature data with the target comment samples to construct a training dataset; The training dataset is input into the preset comment generation model, and the model parameters are optimized by using supervised learning combined with a loss function. The generalization ability of the model is evaluated to complete the model training. The energy consumption-related quantitative data is input into a pre-trained comment generation model, which outputs vehicle energy consumption level results and weighted ranking results of energy consumption influencing factors. A library of pre-defined energy consumption analysis scenarios categorized by vehicle driving conditions, with each scenario associated with a different energy consumption level and a comment template. Based on the energy consumption level and operating condition weight output by the model, a corresponding evaluation template is matched, and quantitative data and the ranking results of energy consumption influencing factors are embedded to generate an energy consumption performance evaluation result.

2. The method for generating vehicle energy consumption performance evaluation based on deep learning according to claim 1, characterized in that, The energy consumption-related quantitative data includes: fuel consumption data, vehicle speed data, road type data, and terrain data. The operating condition proportion feature refers to the proportion of driving mileage corresponding to each type of road and terrain to the total driving mileage.

3. The method for generating vehicle energy consumption performance evaluation based on deep learning according to claim 1, characterized in that, The preset comment generation model is specifically as follows: A model combining long short-term memory networks with attention mechanisms; The model includes: Input layer: Converts the input feature data of the model into multi-dimensional feature vectors and inputs them into the model; Feature extraction layer: processes the input feature data, captures the correlation between data, extracts temporal features from the data, and identifies and analyzes potential patterns; Key Feature Focusing Layer: The output features of the feature extraction layer are weighted and fused, and the weights are dynamically assigned to the energy consumption evaluation based on the features. Feature integration layer: Through linear transformation combined with nonlinear activation function, the weighted and fused features are nonlinearly combined and dimensionally mapped to extract high-dimensional abstract features; Probabilistic output layer: The output of the feature integration layer is converted into a probability distribution using a normalized probabilistic activation function, generating the basis for probabilistic output.

4. The method for generating vehicle energy consumption performance evaluation based on deep learning according to claim 1, characterized in that, The optimization model parameters include: The loss value during model training is calculated using a cross-entropy loss function. The model parameters are adjusted using gradient descent-type optimization algorithms. When the loss value tends to stabilize and the model's generalization ability meets the preset requirements, the model training is completed.

5. The method for generating vehicle energy consumption performance evaluation based on deep learning according to claim 1, characterized in that, include: The vehicle energy consumption rating results are multi-level quantitative ratings; The ranking results of energy consumption influencing factors include the weight percentage and priority relationship of each influencing factor.

6. The method for generating vehicle energy consumption performance evaluation based on deep learning according to claim 1, characterized in that, include: The classification based on vehicle driving conditions specifically refers to the combination of road type and terrain distribution. The comment template reserves placeholders for quantitative information. The quantitative information refers to numerical data related to energy consumption, operating condition ratio, and driving behavior.

7. The method for generating vehicle energy consumption performance evaluation based on deep learning according to claim 1, characterized in that, The generated energy consumption performance evaluation of the commercial vehicle includes: The content includes energy consumption levels, driving conditions, quantitative statistics, factors affecting energy consumption, and optimization suggestions.

8. A vehicle energy consumption performance evaluation and generation device based on deep learning, characterized in that, include: The system includes modules for data acquisition and extraction, data preprocessing, model building, dataset building, model training, energy consumption analysis, scenario library, and evaluation result generation. The data acquisition and extraction module is used to collect vehicle operation data and extract energy consumption-related quantitative data from the vehicle operation data. The data preprocessing module is used to statistically analyze the operating condition proportion characteristics of each data item, and then performs noise and outlier cleaning and dimension normalization on the statistically analyzed data to obtain the feature data for model input. The model building module is used to build a preset comment generation model; The dataset building module is used to obtain target comment samples; The preprocessed data is associated with the target comment samples to construct a training dataset; The model training module is used to input the training dataset into the preset comment generation model, optimize the model parameters by using supervised learning combined with a loss function, and evaluate the model's generalization ability through verification methods to complete the model training. The energy consumption analysis module is used to input the energy consumption-related quantitative data into a pre-trained preset comment generation model, and output the vehicle energy consumption level results and the weighted ranking results of energy consumption influencing factors. The scenario library module is used to preset energy consumption analysis scenario libraries divided according to vehicle driving conditions, and each scenario is associated with different energy consumption level evaluation templates. The evaluation result generation module is used to match the corresponding comment template according to the energy consumption level and operating condition weight output by the model, embed quantitative data and the ranking results of energy consumption influencing factors, and generate energy consumption performance evaluation results.

9. An electronic device, characterized in that, include: The processor, communication interface, memory, and communication bus are connected, with the processor, communication interface, and memory communicating with each other via the communication bus. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the deep learning-based vehicle energy consumption performance evaluation generation method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, It stores a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of a deep learning-based vehicle energy consumption performance evaluation generation method as described in any one of claims 1-7.

Images